The Case for Boycotting Generative AI

Key Points:

• Leading AI researchers and insiders, including an Anthropic employee and OpenAI’s chief scientist, have publicly acknowledged the irreversible and dangerous risks posed by current generative AI systems, highlighting the lack of effective risk mitigation or alignment solutions.
• The core issue is not AI being "too smart," but its unreliability and unpredictability, especially with large language models (LLMs) that are probabilistic rather than deterministic, making them difficult to trust and control when given extensive real-world power and internet access.
• There is a call for an immediate pause or boycott on the development and deployment of "incorrigible AI"—AI systems that cannot be reliably aligned with human values—until technical and political frameworks ensure safety and trustworthiness.
• The government’s response has been inadequate, with opaque policies and close ties to industry hindering meaningful regulation, while geopolitical concerns, such as competition with China, should instead motivate international agreements to manage AI risks collaboratively.
• A public boycott of generative AI tools is proposed as a practical measure to pressure AI companies economically, forcing them to prioritize safety over rapid, risky development, as current commercial incentives do not justify the potential catastrophic risks involved.